Multiple pathways impact the swarming motility of Pseudomonas fluorescens Pf0-1

ABSTRACT Swarming motility in pseudomonads typically requires both a functional flagellum and the production/secretion of a biosurfactant. Published work has shown that the wild-type Pseudomonas fluorescens Pf0-1 is swarming deficient due to a point mutation in the gacA gene, which until recently was thought to inactivate rather than attenuate the Gac/Rsm pathway. As a result, little is known about the underlying mechanisms that regulate swarming motility by P. fluorescens Pf0-1. Here, we demonstrate that a ΔrsmA ΔrsmE ΔrsmI mutant, which phenotypically mimics Gac/Rsm pathway overstimulation, is proficient at swarming motility. RsmA and RsmE appear to play a key role in this regulation. Transposon mutagenesis of the ΔrsmA ΔrsmE ΔrsmI mutant identified multiple factors that impact swarming motility, including pathways involved in flagellar synthesis and biosurfactant production/secretion. We find that loss of genes linked to biosurfactant Gacamide A biosynthesis or secretion impacts swarming motility, as does loss of the alternative sigma factor FliA, which results in a defect in flagellar function. Collectively, these findings provide evidence that P. fluorescens Pf0-1 can swarm if the Gac/Rsm pathway is activated, highlight the regulatory complexity of swarming motility in this strain, and demonstrate that the cyclic lipopeptide Gacamide A is utilized as a biosurfactant for swarming motility. IMPORTANCE Swarming motility is a coordinated process that allows communities of bacteria to collectively move across a surface. For P. fluorescens Pf0-1, this phenotype is notably absent in the parental strain, and to date, little is known about the regulation of swarming in this strain. Here, we identify RsmA and RsmE as key repressors of swarming motility via modulating the levels of biosurfactant production/secretion. Using transposon mutagenesis and subsequent genetic analyses, we further identify potential regulatory mechanisms of swarming motility and link Gacamide A biosynthesis and transport machinery to swarming motility. Swarming motility is a coordinated process that allows communities of bacteria to collectively move across a surface. For P. fluorescens Pf0-1, this phenotype is notably absent in the parental strain, and to date, little is known about the regulation of swarming in this strain. Here, we identify RsmA and RsmE as key repressors of swarming motility via modulating the levels of biosurfactant production/secretion. Using transposon mutagenesis and subsequent genetic analyses, we further identify potential regulatory mechanisms of swarming motility and link Gacamide A biosynthesis and transport machinery to swarming motility.

Gac/Rsm pathway could be stimulated by the overproduction of the upstream histidine kinase GacS, suggesting that the N109P mutation in the gacA gene attenuates rather than abolishes the Gac/Rsm pathway of P. fluorescens Pf0-1 and that overstimulation of the pathway could be achieved via chromosomal deletions of rsmA, rsmE, and rsmI genes (36).The loss of these rsm genes results in overstimulation because the respective Rsm proteins are not present to bind and regulate the translation of their target mRNAs (37,38).
Here we show that the ΔrsmA ΔrsmE ΔrsmI triple mutant of P. fluorescens Pf0-1 is capable of swarming motility.We demonstrate that loss of RsmA and RsmE functions is sufficient to induce swarming by P. fluorescens Pf0-1.Starting with the ΔrsmA ΔrsmE ΔrsmI mutant, we use a genetic screen and subsequent molecular genetic studies to identify pathways involved in swarming motility.These pathways impact flagellar motility and link the biosynthesis of the biosurfactant Gacamide A and its secretion machinery with swarming motility.

The Gac/Rsm pathway regulates swarming motility in a RsmA-and RsmEdependent manner
To assess the role of the Gac/Rsm pathway for swarming motility, we utilized a strain deficient in the small proteins RsmA, RsmE, and RsmI (Rsm proteins) in P. fluorescens Pf0-1, which was previously demonstrated to mimic overstimulation of the Gac/Rsm Pathway (36).We probed this strain for flagellar function using a Tris-buffered mini mal medium supplemented with L-arginine as the sole carbon source (KA medium) supplemented with 0.3% agar (designated "swim agar" for assessing swimming motility) or 0.5% agar (designated "swarm agar" for assessing swarming motility).
As previously shown, low percentage agar media, including swarm agar, can also be used to probe for biosurfactant production/secretion by pseudomonads.Biosurfactant production is phenotypically defined as a translucent liquid zone radiating from the inoculum that then forms the leading edge of the subsequent swarm zone (11,39,40).In this study, we designate this phenotype on swarm agar as the "biosurfactant zone." As a negative control, we included a strain deficient in FleQ function (a ΔfleQ mutant), which has previously been shown to lack flagellar motility in multiple pseudomonads (24,27,(41)(42)(43)(44).
After 24-h growth on swim agar at 30°C, the ΔrsmA ΔrsmE ΔrsmI triple mutant, which was capable of swimming, showed significantly less swimming motility compared to the wild-type (WT) strain (Fig. 1A).This reduced swimming phenotype was reported previously by Pastora and O'Toole (36) and attributed to differences in planktonic growth (36).By contrast, the ΔfleQ and ΔfleQ ΔrsmA ΔrsmE ΔrsmI mutant strains were completely deficient for swimming motility (Fig. 1A).
The ΔrsmA ΔrsmE ΔrsmI mutant produced a biosurfactant zone while the WT strain and FleQ-deficient mutants did not produce a measurable biosurfactant zone after 24-h growth on swarm agar at 30°C (Fig. 1B).Interestingly, the ΔrsmA ΔrsmE ΔrsmI mutant also produced a sizeable but smaller biosurfactant zone when grown for 24 h at room temperature (Fig. S1A).Since the initial 24-h incubation time was not sufficient to induce swarming motility in any of the tested strains or temperature conditions, we subse quently incubated the strains for an additional 24 h at room temperature.After 24-h growth on swarm agar at 30°C and then 24-h growth at room temperature, the ΔrsmA ΔrsmE ΔrsmI mutant displayed motility on swarm agar, whereas the other strains were deficient for swarming motility (Fig. 1C).The ΔrsmA ΔrsmE ΔrsmI mutant produced a similar swarm zone when grown on swarm agar for 48 h at room temperature but produced a significantly smaller swarm zone when grown on swarm agar for 48 h at 30°C (Fig. S1B).Interestingly, the ΔrsmA ΔrsmE ΔrsmI mutant swims, produces biosurfactant, and eventually swarms on swim agar (Fig. S2), although to a lesser extent than on swarm agar.
To identify the Rsm proteins associated with swarming motility, we made chromoso mal deletions of the individual genes coding for the Rsm proteins in the WT and FleQdeficient background strains and assessed these strains on swarm agar.Loss of any individual Rsm protein was not sufficient to induce swarming motility (Fig. 2A).Analysis of these strains for swimming motility and biosurfactant production revealed that while the single rsm deletions in the WT background were proficient for swimming motility (Fig. S3A), none of these strains were able to produce a biosurfactant zone (Fig. S3B).
We made combinatorial chromosomal deletions of the genes coding for the various Rsm proteins in the WT and FleQ-deficient backgrounds and assessed these mutants for their ability to swarm.After growth for 24 h at 30°C and then 24 h at room temperature, the ΔrsmA ΔrsmE double mutant was able to swarm to levels similar to the ΔrsmA ΔrsmE ΔrsmI triple mutant, while the other combinatorial deletions were deficient for swarming motility (Fig. 2B).Analysis of these strains for swimming motility and biosurfactant production revealed that all deletions in the WT background were proficient for swim ming motility (Fig. S4A), but of these strains, only the ΔrsmA ΔrsmE double and ΔrsmA ΔrsmE ΔrsmI triple mutants were proficient for biosurfactant production (Fig. S4B).
To further demonstrate the role of RsmA and RsmE in regulating swarming motility, we inserted the coding sequence of RsmA or RsmE into the arabinose-inducible shuttle vector pMQ72 directly downstream of the P BAD promoter (see Materials and Methods), transformed the constructs into the ΔrsmA ΔrsmE ΔrsmI and ΔfleQΔrsmA ΔrsmE ΔrsmI backgrounds and assessed these strains for their ability to swarm.After growth for 24 h at 30°C and then 24 h at room temperature on swarm agar without inducing conditions (no addition of arabinose), the ΔrsmA ΔrsmE ΔrsmI + pMQ72 rsmA and ΔrsmA ΔrsmE ΔrsmI + pMQ72 rsmE strains are completely deficient for swarming motility (Fig. 2C and  D, respectively).Analysis of these strains revealed that ΔrsmA ΔrsmE ΔrsmI + pMQ72 rsmA and ΔrsmA ΔrsmE ΔrsmI + pMQ72 rsmE strains were proficient in swimming motility (Fig. S5A and B), and these strains had a statistically significant decrease in biosurfactant secretion compared to their parental strains (Fig. S5C and D).

Transposon mutagenesis reveals genes required for swarming motility
Given our observation that the ΔrsmA ΔrsmE ΔrsmI strain is proficient for swarming motility, we generated a TnM mariner transposon mutant library as previously described in Pastora and O'Toole (36).We initially probed this library for candidates that were able to swarm after 24 h of growth at 30°C on swarm agar (hyper-swarm candidates) or were completely deficient in swarming motility after 24 h of growth at 30°C and then 24 h of growth at room temperature on swarm agar (swarm-deficient candidates) and identified the approximate transposon insertion site and P TAC promoter orientation using arbitrary primed PCR.
These candidates were re-screened for swimming motility, biosurfactant production, and swarming motility.All candidates are listed in Table S1 (n =108) with the associated locus tag, gene name, GenBank description, KEGG BRITE Terms, and KEGG Pathway Terms (where applicable).The candidates were also grouped separately into hyper-swarm candidates (Table S2, n = 18) and swarm-deficient candidates (Table S3, n = 90).The swarm-deficient candidates were additionally grouped independently based on loss of biosurfactant production (Table S4, n = 25) or loss of swimming motility (Table S5, n = 82; note some of the candidates are listed in more than one table).The candidates in each table were categorized based on annotated function and are graphically summarized in Fig. 3. rsmA after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.(D) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI mutant, ΔrsmA ΔrsmE ΔrsmI mutant + pMQ72 rsmE, and ΔfleQ ΔrsmA ΔrsmE ΔrsmI mutant + pMQ72 rsmE after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.Statistical significance for this figure was determined using one-way ANOVA with Tukey's multiple comparisons tests.****P < 0.0001.All error bars represent standard deviation.

Loss of genes related to Gacamide A biosynthesis and secretion result in loss of swarming motility and changes in biosurfactant production
We identified a class of mutants defective for both swarming motility and biosurfactant production (Table S4) but swim proficiently.These mutants all had transposon insertions that mapped to a biosynthetic operon reported to encode the machinery needed to produce the non-ribosomal cyclic lipopeptide biosurfactant Gacamide A. Briefly, this machinery consists of modules containing a condensation, adenylation, and thiolation domain which iteratively add an amino acid to the growing peptide chain.The growing chain is then released and circularized by the tandem thioesterase domain encoded in gamC (45).Gacamide A can promote swarming motility when purified and exogenously supplemented to the swarm-deficient parental P. fluorescens Pf0-1 (34,45,46).The gamA, gamB, and gamC genes encode the biosynthetic proteins for Gacamide A. The pleA, pleB, and pleC genes are predicted to encode the Gacamide A secretion system given their sequence similarity to the MacAB-TolC type macrolide efflux pumps and conservation among cyclic lipopeptide-producing pseudomonads (45,(47)(48)(49).
Given these identified candidate genes, we conducted a genetic analysis utilizing the swarm-proficient ΔrsmA ΔrsmE ΔrsmI strain and introduced additional single chromo somal deletions of the predicted biosynthetic (gamA, gamB, and gamC) genes and transport (pleA, pleB, and pleC) genes.For these operonic genes, the individual deletion constructs were designed such that the relevant promoters remained intact to drive the expression of any downstream genes.These strains were then probed for biosurfactant production and swarm motility.For the genes predicted to encode functions required for Gacamide A production, deletion of gamA resulted in a marked decrease in biosurfactant production (Fig. 4A) and swarming motility (Fig. 4B), deletion of gamB resulted in a modest, but statistically significant decrease in biosurfactant production (Fig. 4C) and swarming motility (Fig. 4D), and deletion of gamC resulted in a complete loss of biosurfactant production (Fig. 4E) and swarming motility (Fig. 4F).Representative images of the swarm zones are included in Fig. S6.These results agree with previous work focused on gamA in a P. fluorescens Pf0-1 merodiploid strain containing both the native and Pseudomonas protogens Pf-5 gacA allele (34).For the genes predicted to encode functions required for Gacamide A export, loss of the predicted periplasmic or outer membrane components encoded by pleA and pleC gene, respectively, resulted in the complete loss of the biosurfactant zone (Fig. 5A and E, respectively) and swarming motility (Fig. 5B and F, respectively).Interestingly, loss of the predicted inner membrane component encoded by pleB did not result in a reduction of the biosurfactant zone (Fig. 5C) but did result in a sizeable decrease in swarming motility (Fig. 5D).Representative images of the swarm zones are included in Fig. S7.
To confirm that the observed differences in swarming motility were not related to changes in flagellar function, we assessed the strains for their ability to swim.After 24 h of growth at 30°C on swim agar, all tested mutants produced swim zones similar to their parental strain (Fig. S8 and S9).The similar swim zones also indicated that there was no general growth defect for any of these mutants.

Loss of FliA results in a swarming and swimming motility defect
We identified several candidate mutants defective for both swimming and swarming motility (Table S5) but biosurfactant proficient.We focused our analysis specifically on the alternative sigma factor FliA, which was previously associated with flagellar biosynthesis and swarming motility regulation in P. aeruginosa (50,51).We made a chromosomal deletion of fliA in the swarm-proficient ΔrsmA ΔrsmE ΔrsmI strain and assessed the mutant for the ability to swim and swarm.
After 24 h of growth at 30°C on swim agar, the ΔrsmA ΔrsmE ΔrsmI ΔfliA strain was completely deficient in swimming motility (Fig. 6A).After 24 h of growth at 30°C and 24 h of growth at room temperature on swarm agar, the ΔrsmA ΔrsmE ΔrsmI ΔfliA strain was completely deficient in swarming motility (Fig. 6B).We noted no significant difference in biosurfactant production between the ΔrsmA ΔrsmE ΔrsmI ΔfliA mutant and its parental strain (Fig. S10), indicating that loss of FliA function results in loss of motility via loss of flagellar function.

DISCUSSION
Our data here build upon the link between the Gac/Rsm pathway and swarming motility reported by our group and several other teams (22,32,34,35,42,45,46).Our genetic analyses demonstrate that loss of the Rsm proteins, which was previously shown to phenotypically mimic overstimulation of the Gac/Rsm pathway (36), induces biosurfac tant production/secretion and swarming motility of P. fluorescens Pf0-1 and does so in a RsmA-and RsmE-dependent manner.Interestingly, previously published work demon strated that loss of rsmE was sufficient to induce biosurfactant production on nutrientrich Pseudomonas F Agar overlaid with polycarbonate membrane (46).However, this phenotype was not observed in this study when the rsmE-deficient strain was grown on swarm agar.We posit that these phenotypic differences are due to the media differences between these two studies, as Evans et al. optimized their conditions specifically for biosurfactant production (46), while the current study utilized a minimal, low-agar medium for swarming motility.Despite these differences, this current study demon strates that overexpression of RsmE is sufficient to repress biosurfactant production and swarming motility.These findings support previously published data showing that a P. fluorescens Pf0-1 merodiploid strain containing the native Pf0-1 gacA gene and non-native gacA gene from P. protogens Pf-5 was proficient for swarming motility (34) and build on previous results demonstrating that RsmE regulates biosurfactant secretion by P. fluorescens Pf0-1 (46).
Interestingly, variation of the incubation profile suggests that biosurfactant deployment and swarming motility are both temperature-dependent processes.
Contradictorily, biosurfactant deployment was optimal at 30°C while swarming motility was optimal at room temperature.These associations between temperature, motility, and biosurfactant deployment can be explored in future studies.
Our subsequent transposon mutagenesis identified numerous genes that contrib ute to swarming motility via the production of a functional flagellum, biosurfactant biosynthesis/secretion, or both.The mutants with a defect in biosurfactant production ΔpleB quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleB quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.(E) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔpleC quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleC quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(F) Swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔpleC quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleC quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.Statistical significance was determined using one-way ANOVA with Tukey's multiple comparisons tests.****P < 0.0001.All error bars represent standard deviation.included strains with transposon insertions in genes encoding the non-ribosomal peptide synthetases responsible for synthesizing the biosurfactant Gacamide A and its predicted outer membrane transporter PleC.Subsequent genetic analyses revealed that loss of gamA, gamB, or gamC decreased biosurfactant secretion and swarming motility to varying degrees.These differences could be attributed to the likelihood that loss of the gamA or gamB gene still results in the production of basal levels of Gacamide A or an alternative surfactant that can partially support swarming motility.Evidence supporting this second idea has been previously reported by other groups, whereby domain deletions and substitutions within non-ribosomal peptide synthetases result in an alternative peptide being synthesized, which is a strategy currently being utilized to develop novel biotherapeutics (52)(53)(54)(55)(56)(57).Based on the observed complete loss of swarming motility but the ability of this strain to swim, a mutation in the gamC gene likely results in a complete loss of biosurfactant production.The predicted terminal tandem thioesterase (TE-TE) domain of GamC is likely required for the circularization and release of the synthesized peptide from its peptidyl carrier protein (45,58,59).Interestingly, previously published mutations in gamA via antibiotic resistance cassette (34) or mini-Tn5 insertion (46) resulted in a complete loss of biosurfactant secretion, whereas a chromosomal deletion resulted in a sizeable decrease but not complete loss of biosurfactant secretion.We posit that disruption of the gamA gene via inser tional mutagenesis has the unintended consequence of inactivating the whole operon, whereas chromosomal deletion likely allows for the continued production of the GamB and GamC proteins.With the loss of GamA via chromosomal deletion, it is likely that the peptide synthesis is initiated at the first condensation domain within the first module encoded in GamB resulting in a truncated variant of Gacamide A.
Genetic analysis revealed that the predicted Gacamide A secretion machinery is also required for swarming motility.As expected, loss of the periplasmic component PleA or the outer membrane component PleC of the proposed Gacamide A secretion machinery resulted in loss of detectable biosurfactant and complete loss of swarming motility.Interestingly, loss of the inner-membrane component PleB results in a complete loss of swarming motility even though this mutant strain is apparently proficient for biosurfac tant production and swimming motility.These observations could suggest the presence of an alternative inner membrane transporter component that can substitute for PleB to support the secretion of Gacamide A, or that PleB may modify Gacamide A during secretion, which may be required for this surfactant to promote swarming motility.
The identified candidates with defects only in swimming and swarming motility clustered to a variety of seemingly unrelated pathways.Since multiple studies will be required to fully investigate the role of these many functions in motility, we focused here specifically on the alternative sigma factor FliA.As expected based on other studies in pseudomonads (50,51,60), a loss of FliA resulted in the loss of swimming and swarming motility, likely via loss of flagellar function.
Collectively, our findings provide evidence that swarming motility is regulated via the Gac/Rsm pathway in a RsmA-and RsmE-dependent manner by impacting flagellar motility and biosurfactant production/secretion.These findings provide a set of tools for the future study of these pathways, and we hope the findings from the transposon screen may be used by the broader community to study both motility and other Gac/ Rsm-dependent phenotypes of P. fluorescens Pf0-1.

Strains and media used in this study
P. fluorescens Pf0-1, E. coli S17-1 λ-pir, and E. coli SM10 λ-pir were used throughout this study.E. coli and P. fluorescens were routinely grown in lysogeny broth (LB) and P. fluorescens was routinely grown in KA minimal medium, as previously defined by Collins et al. (61).KA minimal medium contains 50 mM Tris-HCl (pH 7.4), 0.61 mM MgSO 4 , 1 mM K 2 HPO 4 , and 0.4% (wt/vol) L-arginine HCl.The medium was supplemented with 30 µg/mL gentamycin for P. fluorescens when harboring the expression vector pMQ72.For E. coli, the medium was supplemented with 10 ug/mL gentamycin when harboring the allelic exchange vector pMQ30 or the expression vector pMQ72, with 50 µg/mL carbenicillin for the strain harboring the transposon-containing shuttle vector pBT20, or 15 ug/mL tetracycline when harboring the allelic exchange plasmid pEX18Tc.The strains and plasmids used in this study are listed in Table S6.

Construction of in-frame chromosomal gene deletions
For chromosomal deletions in this study, the allelic exchange vectors pEX18Tc or pMQ30 were utilized.Flanking regions of the target genes were amplified via PCR using Phusion polymerase (New England BioLabs).All primers used in the study are listed in Table S7.
The amplicons were integrated into SmaI (New England BioLabs)-digested vector using the GeneArt Gibson Assembly HiFi Master Mix (Invitrogen) according to the manufacturer's specifications.Constructs were electroporated into E. coli S17-1 λ-pir and plated on LB agar with 15 µg/mL tetracycline or 10 µg/mL gentamycin for pEX18Tc or pMQ30-based constructs, respectively.Integration of the flanking regions into the vector was confirmed via PCR amplification and Sanger sequencing.Constructs were conjugated into P. fluorescens, whereby 1 mL aliquots of E. coli harboring the construct and P. fluorescens cultures grown overnight at 30°C were mixed in a 2 mL microcentrifuge tube, pelleted, washed in LB, and then plated on LB agar with no antibiotic selection to facilitate uptake of the deletion construct by P. fluorescens.After overnight incubation at 30°C, the cells were scraped from the plate and resuspended in fresh LB liquid medium.Serial dilutions were plated on LB agar supplemented with 45 µg/mL tetracycline and 30 µg/mL chloramphenicol for pEX18TC-based constructs or 30 µg/mL gentamycin and 30 µg/mL chloramphenicol for pMQ30-based constructs to select and counter select, respectively, for integration of the constructs into the P. fluorescens genome.To facilitate looping out of the drug resistance cassette, merodiploid candidates were initially grown in LB liquid without antibiotic selection overnight at 30°C and then serial dilutions were plated on LB agar without sodium chloride supplemented with 10% sucrose.Single colonies were struck on LB agar with and without antibiotic selection to confirm the loss of the resistance cassette carried by the plasmid.Antibiotic-susceptible candidates were screened for loss of the target gene via PCR amplification and Sanger sequencing.

Construction of the rsmA expression vector
The rsmA gene was PCR amplified using Phusion polymerase (New England BioLabs) according to the manufacturer's specifications.The primers were designed such that the forward primer contained the high-affinity T7 phage gene 10 ribosomal binding sites (62) and 8 bp spacer nucleotides upstream of the rsmA start codon.In addition, both primers contained 20 bp of homology to the pMQ72 SmaI cut-site at their 5′ ends.Primers used to build this construct are listed in Table S7.
The amplicon was integrated into SmaI-digested pMQ72 using the GeneArt Gibson Assembly HiFi Master Mix (Invitrogen) according to the manufacturer's specifications.The construct was electroporated into E. coli S17-1 λ-pir, cells were recovered in LB for 1 h at 30°C and subsequently plated on LB agar supplemented with 10 µg/mL gentamycin.Antibiotic-resistant candidates were sequenced to confirm gene insertion into pMQ72.The construct was then electroporated into P. fluorescens, cells were initially recovered in LB for 1 h at 30°C, and dilutions were plated on LB agar supplemented with 30 µg/mL gentamycin to select for retention of the construct.

Swim assay
Swim assays were conducted as previously defined in Pastora and O'Toole (36).Briefly, P. fluorescens Pf0-1 strains were grown overnight in 5 mL of LB with appropriate antibiotic selection at 30°C with agitation.1 mL of aliquots was transferred to sterile 1.5 mL microcentrifuge tubes and used to stab inoculate KA minimal medium supplemented with 0.3% agar (swim agar).Inoculated plates were incubated for 24 h at 30°C.The diameter of the resulting swim zones was measured using a ruler.

Biosurfactant and swarm assays
Biosurfactant and swarm assays were conducted using a modified approach previously described by Ha et al. (63).Briefly, P. fluorescens Pf0-1 strains were grown overnight in 5 mL of LB with appropriate antibiotic selection at 30°C with agitation.2.5 µL of each culture was pipetted directly on the surface of KA minimal medium supplemented with 0.5% agar (swarm agar).Plates were initially incubated for 24 h at 30°C and the diameter of the resulting biosurfactant zones was measured using a ruler.Plates were then incubated for an additional 24 h at room temperature and the diameter of the resulting swarm zones was measured using a ruler.

Transposon mutagenesis
Transposon mutagenesis was conducted as previously defined in Pastora and O'Toole (36).Briefly, E. coli harboring the pBT20 plasmid and the P. fluorescens Pf0-1 ΔrsmAΔrsmEΔrsmI triple mutant were grown in LB, with appropriate antibiotic selection, overnight at 30°C, and mixed in a 1:1 ratio in a 2 mL microcentrifuge tube.Cell pellets were then washed, resuspended, and plated on LB agar to facilitate conjugation of the pBT20 plasmid into P. fluorescens.Plates were incubated for 90 min at 30°C, and the resulting cells were scraped from the plate, washed in fresh LB, serially diluted, and plated on LB agar supplemented with 30 µg/mL gentamycin and 30 ug/mL chloramphe nicol to select for transposon integration in P. fluorescens.Per Pastora and O'Toole (36), mutant libraries were generated by picking single colonies with sterile pipette tips into sterile 96-well flat-bottom polystyrene plates.Candidates were subsequently screened for their ability to swarm on swarm agar (see above).Candidates that were unable to swarm or were able to swarm at the 24-h timepoint and continue to swarm at the 48-h timepoint (hyper-swarmers) were identified from the mutant library.The identified candidates were struck out from the mutant library onto LB agar supplemented with 30 µg/mL gentamycin and subsequently re-tested for swarming motility to verify the swarm-deficient or hyper-swarm phenotype.Verified candidates were subsequently phenotypically screened for swimming motility and biosurfactant secretion.Transpo son chromosomal insertion sites and transposon orientations were determined using arbitrary primed PCR as described by O'Toole et al. (64).

FIG 1 A
FIG 1 A strain lacking the Rsm proteins swarms.(A) Swim zone (in millimeters) of the WT strain, ΔfleQ single mutant, ΔrsmA ΔrsmE ΔrsmI triple mutant, and a ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant after toothpick inoculation on KA minimal medium supplemented with 0.3% agar followed by 24-h growth at 30°C.(B) The biosurfactant zone (in millimeters) of the WT strain, ΔfleQ single mutant, ΔrsmA ΔrsmE ΔrsmI triple mutant, and a ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(C) Swarm zone (in millimeters) of the WT strain, ΔfleQ single mutant, ΔrsmA ΔrsmE ΔrsmI triple mutant, and a ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C then 24-h growth at room temperature.Statistical significance for this figure was determined using one-way ANOVA with Tukey's multiple comparisons tests.****P < 0.0001.All error bars represent standard deviation.

FIG 3
FIG 3 Summary of genetic loci identified by transposon mutagenesis.Stacked bar charts categorizing the candidates (n = 108) identified from the transposon mutagenesis.Candidates were phenotypically sub-grouped based on swarming motility into hyper-swarming (n = 18) or swarm deficient (n = 90).The swarm-deficient group was then sub-grouped based on a loss of biosurfactant production (n = 25) or a defect in swimming motility (n = 82), with a subset of candidates present in both groups.

FIG 4
FIG 4 Loss of components of the biosurfactant operon effect swarming motility through the altered biosurfactant levels.(A) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamA quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(B) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamA quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.(C) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamB quadruple mutant, andthe ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamB quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(D) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamB quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamB quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.(E) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamC quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamC quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(F) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔgamC quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔgamC quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.Statistical significance for this figure was determined using one-way ANOVA with Tukey's multiple comparisons tests.****P < 0.0001.All error bars represent standard deviation.

FIG 5
FIG 5 Loss of any component of the biosurfactant secretion system results in a swarm motility defect.(A) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔpleA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleA quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(B) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔpleA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleA quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% followed by 24-h growth at 30°C and then 24-h growth at room temperature.(C) The biosurfactant zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔpleB quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔpleB quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C.(D) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI

FIG 6 A
FIG 6 A FliA-deficient strain has a motility defect likely due to loss of flagellar function.(A) The swim zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔfliA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔfliA quintuple mutant after toothpick inoculation on KA minimal medium supplemented with 0.3% agar followed by 24-h growth at 30°C.(B) The swarm zone (in millimeters) of the ΔrsmA ΔrsmE ΔrsmI triple mutant, ΔfleQ ΔrsmA ΔrsmE ΔrsmI quadruple mutant, ΔrsmA ΔrsmE ΔrsmI ΔfliA quadruple mutant, and the ΔfleQ ΔrsmA ΔrsmE ΔrsmI ΔfliA quintuple mutant after inoculation of 2.5 µL of overnight culture on the surface of KA minimal medium supplemented with 0.5% agar followed by 24-h growth at 30°C and then 24-h growth at room temperature.Statistical significance for this figure was determined using one-way ANOVAs with Tukey's multiple comparisons tests.****P < 0.0001.All error bars represent standard deviation.